Look-Back and Look-Forward Durations and the Apparent Appropriateness of Ambulatory Antibiotic Prescribing

Abstract

Ambulatory antibiotic stewards, researchers, and performance measurement programs choose different durations to associate diagnoses with antibiotic prescriptions. We assessed how the apparent appropriateness of antibiotic prescribing changes when using different look-back and look-forward periods. Examining durations of 0 days (same-day), −3 days, −7 days, −30 days, ±3 days, ±7 days, and ±30 days, we classified all ambulatory antibiotic prescriptions in the electronic health record of an integrated health care system from 2016 to 2019 (714,057 prescriptions to 348,739 patients by 2391 clinicians) as chronic, appropriate, potentially appropriate, inappropriate, or not associated with any diagnosis. Overall, 16% percent of all prescriptions were classified as chronic infection related. Using only same-day diagnoses, appropriate, potentially appropriate, inappropriate, and not-associated antibiotics, accounted for 14%, 36%, 22%, and 11% of prescriptions, respectively. As the duration of association increased, the proportion of appropriate antibiotics stayed the same (range, 14% to 18%), potentially appropriate antibiotics increased (e.g., 43% for −30 days), inappropriate stayed the same (range, 22% to 24%), and not-associated antibiotics decreased (e.g., 2% for −30 days). Using the longest look-back-and-forward duration (±30 days), appropriate, potentially appropriate, inappropriate, and not-associated antibiotics, accounted for 18%, 44%, 20%, and 2% of prescriptions, respectively. Ambulatory programs and studies focused on appropriate or inappropriate antibiotic prescribing can reasonably use a short duration of association between an antibiotic prescription and diagnosis codes. Programs and studies focused on potentially appropriate antibiotic prescribing might consider examining longer durations.

Prevalence and appropriateness of in-person versus not-in-person ambulatory antibiotic prescribing in an integrated academic health system: A cohort study

“…Additionally, we compiled all diagnosis codes coded by the antibiotic prescriber up to 21 days later, as our manual chart review revealed instances in which a clinician would initiate an encounter, apply a diagnosis, and, within that same encounter, prescribe an antibiotic up to 21 days later. A subsequent analysis with the data from the present study revealed subtle differences in results using different durations between the antibiotic prescription and diagnosis codes [ 26 ].…”